Large Eddy Simulations of a Mach 2 Expansion Compression Corner

High-speed flow over an expansion compression corner includes many flow features that prove challenging to accurately predict with turbulence models. Recent research into machine learning corrections and turbulence models have shown promise for improving Reynolds Averaged Navier Stokes predictions, however these methods require a wide range of training datasets in order to be generalized for engineering purposes. To aid in the generation of usable datasets, large eddy simulations of a Mach 2 expansion compression geometry are presented. The same geometry is tested at 2 different friction Reynolds numbers of 500 and 1040. It is shown that both cases experience relaminarization. The lower Reynolds number case shows earlier onset of separation, while the higher Reynolds number case shows signs of retransition prior to the onset of separation. Due to this behavior RANS predictions for these cases show delayed onset of separation and earlier onset of separation for the low and high Reynolds number case respectively.